PodBrowser
The Rollup

Matt Hougan & Sal Ternullo: NEAR Is One Of The Best Positioned Assets In Crypto

Tuesday, 6 October 2026 · 3 min read · Listen to the episode ↗

Matt Hougan and Sal Ternullo make the case that NEAR is unusually well positioned because NEAR Intents has demonstrated usage, revenue growth, and signs of product-market fit, giving investors a concrete business to assess alongside its AI ambitions. They discuss private inference as an early signal of demand and NEAR’s recent repricing, which Hougan calls a “down payment” if the long-term thesis proves right.

Matt Hougan sees NEAR as easier to evaluate than many crypto assets because NEAR Intents already has usage, revenue growth, and signs of product-market fit. Its cross-chain interaction product gives investors a business they can assess before deciding whether to underwrite NEAR’s longer-term AI ambitions. Hougan believes that combination leaves NEAR among crypto’s best-positioned assets, though he said awareness of its fit with AI remains limited. Ternullo likewise argued that NEAR had been fundamentally mispriced relative to Intents’ traction and revenue growth.

Ternullo described NEAR’s move over the preceding 30–45 days as the repricing he had expected. Hougan said the rise gives traditional investors a more legible long-term chart and a basis for comparing NEAR’s valuation with its prospects. If his outlook proves right, he called the move a “down payment,” not the full realization of the thesis. The investment case depends on more than a narrative: Intents’ demonstrated traction is the current evidence, while AI represents additional potential. The difficulty, Hougan said, is explaining both parts of NEAR’s story quickly.

Ternullo framed NEAR’s AI thesis around privacy, verifiability, and confidentiality for infrastructure supporting open models. With private inference, clients can verify that operators cannot see prompts, outputs, or related telemetry. Token consumption rose sharply after private inference went live on OpenRouter, which Ternullo characterized as an early sign of product-market fit. The Goose consumer app is intended to bring NEAR features together in a simpler experience; private inference and Iron Claw are meant to demonstrate agentic interactions. Ternullo expects multiple product verticals to scale around one token economy, but said a NEAR repricing would depend on reaching sufficient scale and enabling fee switches, including on the AI side.

NEAR’s token economics have also begun to change as the project has sought to connect network activity with value for holders. Ternullo said inflation fell from 5% to 2.5% on October 28 of the previous year. After NEAR Intents showed stickiness and an integration-side moat, the project began monetizing and returning value through buybacks; he cautioned that buybacks do not necessarily mean tokens are burned. A governance-forum proposal would reduce inflation further, from 2.5% to 1.6% over two years, but it has not been approved. A proposed sovereign wealth fund concept was shelved amid concerns that included discretionary management. Ternullo said NEAR’s tokenomics considers both demand and supply, while Hougan said there is no single proven buyback model and that long-term value accrual to token holders should guide decisions.

Hougan described the broader market as a differentiated bull market in which project-specific developments matter more than a rising tide lifting all assets. Over the prior 30 days, the top 20 assets showed about 130% dispersion between the best and worst performers, in his account. He expects weaker “zombie chains” to fall away and said he believes the market bottomed in June. He pointed to improved regulation, institutional investment, crypto revenue, tokenization, and AI as reasons for that view, rather than presenting it as a certainty. In this environment, NEAR’s case rests on the combination of a functioning, monetizing product and an AI opportunity whose broader recognition and scale remain to be proven.

This summary was generated from the episode transcript and can contain mistakes.